The production of miniaturized electronic devices and circuits, commonly referred to as integrated circuits or "chips", includes a number of uncommon manufacturing techniques. The term "chip" often refers to an integrated circuit product having several layers of metallic, insulating or semiconductor materials which have been deposited on, or removed from, a portion of a semiconducting crystalline silicon substrate, all in some predetermined pattern.
As the manufacture of integrated circuits has developed, including the manufacture of very large scale integrated circuits ("VLSI" circuits), the need for adding more devices in smaller spaces has led to the development of photographic and lithographic techniques in such manufacture. For example, in order to form electronic devices in semiconductor materials, insulator materials, and conductive materials (typically metals), a number of very small and yet very precise geometric patterns ("geometries" ) must be formed in these materials both individually and when they are brought together. Furthermore, production of VLSI circuits typically requires that such patterns extend in three dimensions, specifically between and among various layers of materials.
A present technique for forming geometries in such materials is a patterning technique in which the pattern to be formed in a material is first designed as a mask. A "mask" is a set of images, often fabricated on glass plates and having opaque and transparent parts which represent all, or portions of, the predetermined pattern of semiconductor, metallic or insulating materials, or any combination of these, which is present on, or removed from, a chip.
In many such techniques, the mask is applied to the material to be patterned after the material has been coated with a light sensitive material generally referred to as a photoresist (or sometimes simply a "resist"). As its name implies, the photoresist material is sensitive to light and the material resists particular chemical or physical reactions. This method of patterning is typically termed "contact printing". Even more common today is the use of projection lithography where light is projected through the mask and a lens system onto the wafer surface. Another mechanism for patterning photosensitive materials utilizes electron beam radiation in which a precisely focused beam of electrons is deflected over an electron-beam-sensitive resist film to write patterns without the presence of an intermediate photomask. Still other lithographic methods may employ this resist technology, such as projection of images from masks or reticles and other beam patterning techniques such as laser direct write.
Typical photoresists are solutions of low molecular weight novolac resins, solvents such as propylene glycol methyl ether acetate or 2-ethoxy ethyl acetate, and to which one or more photosensitive compounds such as diazonapthoquinones have been added. When the solvents evaporate, a solid photosensitive film remains. Conventional photoresists may also include dyes to prevent the exposing light from undesirably scattering during exposure, as well as other ingredients such as surfactants.
For example, in "contact printing", the patterned mask is applied over the photoresist and the mask and photoresist are exposed to light. The portions of the photoresist which are left uncovered by the mask undergo photochemical reactions which result in characteristics different from the remainder of the photoresist which were covered by the mask and not exposed to the light. When the mask is removed, the chemical differences between the masked and unmasked portions of the photoresist permit one or the other of those portions to be removed in a specified chemical reaction. For example, in a positive acting resist, exposure to light could render the photoresist soluble in a hydroxide base aqueous solution while nonexposure would leave it insoluble. Thus, developing the photoresist in such a solution would remove the exposed portions and leave the nonexposed portions behind. The result is a layer of material yet to be patterned upon which is a layer of patterned photoresist.
The next step is to pattern the material underneath the photoresist, a common technique for which is plasma etching. As known to those familiar with the manufacture of semiconductor devices, in a plasma etching process, a gas is excited by electron impact to produce ions and reactive neutral species. The generated neutral species diffuse out of the plasma to the substrate. There, they react with the substrate to form volatile products. Activation energy for the volatilization reaction may be supplied by ions which are driven out of the plasma to strike the substrate at an angle near 90.degree.. The directional nature of the activation leads to directionality (anisotropy) in the etched profile; i.e. etching occurs in the direction of ion impact rather than equally in all exposed directions. Plasma etching at low pressure and with a considerable amount of ion activation is called reactive ion etching (RIE). Ideally, the plasma exposure should etch away the exposed portions of the underlying material while leaving the portions covered with photoresist unaffected. When the material has been so patterned, the remaining portions of the photoresist can be removed ("stripped") using an appropriate solvent or solution, leaving behind a patterned semiconductor, insulator, or metallic material.
As the scale of integration of semiconductor devices has become larger, meaning in this sense that a larger number of devices are added to a limited amount of space, additional problems and considerations have arisen. For example, in VLSI chips, the number and density of devices requires patterning of many layers of semiconductor material, insulator material and conductors such as metals. In particular, layers of metals must be carefully insulated from one another and typical materials for such insulation purposes now include synthetic polymers, for example polyimides. As just discussed, such polymer materials must be patterned and etched to produce the geometries appropriate for devices and circuits. Oxygen containing plasmas are typical etch environments for patterning such polymeric insulator materials.
A problem arises, however, when an oxygen containing plasma is used to pattern a polymeric insulator, because photoresist materials useful in patterning, as discussed above, are themselves typically organic polymeric materials. As a result, the oxygen containing plasma often has the same effect on the photoresist as it does upon the organic insulating material being patterned. This means that the photoresist will be etched away at approximately the same rate as the insulator material. This results in many situations in which the photoresist fails to protect the intended portions of the insulator material, resulting in an unsatisfactory device or circuit. As is known to those familiar with such circuits, small imperfections basically render such devices or circuits useless.
Several alternative methods for solving this problem have been attempted. A first attempt has been to make the photoresist proportionally thicker than the organic material being etched. If a substantially equivalent amount of both the organic photoresist and the organic insulator are etched in the oxygen plasma, some of the photoresist should remain protecting the insulator as originally intended. A disadvantage arises, however, because increasing the thickness of the photoresist typically decreases the resolution of the pattern, resulting in a poorer definition of geometries and ultimately in lower quality devices.
The so-called "trilayer techniques" represent a second approach for patterning organic insulators using organic photoresist in oxygen plasmas. In these techniques, the insulator to be etched, such as a polyimide, is first covered with a protective barrier layer of an inorganic material such as silicon dioxide (SiO.sub.2). The photoresist is then added above the inorganic layer. The photoresist is exposed and developed to form a pattern of geometry in the photoresist. This patterned geometry is then transferred through the inorganic layer using a fluorinated plasma such as a carbon tetrafluoride (CF.sub.4) discharge. The pattern formed in the inorganic SiO.sub.2 layer forms a secondary inorganic mask which exposes the complementary pattern in the polyimide. Using the secondary inorganic mask, an oxygen plasma etches the polyimide to the desired geometry, while concurrently removing the remainder of the photoresist. The result is a patterned polyimide covered with an identical pattern of SiO.sub.2. Accordingly, a second etching step such as a CF.sub.4 plasma is required to remove the remaining SiO.sub.2 and obtain the desired geometry in the polyimide alone. In other words, typical barrier techniques require three etching steps, as well as an SiO.sub.2 deposition step in order to pattern a single layer of insulating material. Each processing step to which the substrate is exposed produces more particle related defects and lowers device yield. Conversely, reduction of the number of these steps increases yield. Furthermore, the use of a secondary masking layer impedes the faithful replication of the original mask dimensions and leads to larger dimensional tolerances.
Other techniques incorporate silicon into the photoresist in some fashion in order to resist the effects of the oxygen plasma. Silicon forms an excellent etch barrier in the presence of the oxygen plasma, but lowers the elasticity and adhesion properties of the photoresist. The photoresist then tends to crack, resulting in undesired exposure and patterning where the cracks occur, and often unacceptably alters the desired geometry. Poorer adhesion properties similarly result in delamination in which photoresist layer tends to peel away from the underlying insulating layer. This also has a detrimental result on geometries and resulting devices.